A presentation given to the ESRI NZ User Conference in 2012 about the wind prospecting system developed by Kenex using ArcGIS and custom modelling tools.
Predicting the Wind: Wind farm prospecting using GIS
1. Predicting the wind:
Wind farm prospecting using GIS
Elisa Puccioni
elisa@kenex.co.nz
www.kenex.co.nz
2. Background
• With global energy trends moving towards sustainable solutions, energy explorers are
continuously seeking new ways to effectively search for potential wind farm sites.
• In the past three years, Kenex and Aurecon have developed advanced spatial modelling
techniques that combine:
• Mesoscale modelling of the wind resource to assess the potential of an area
• Advanced terrain analysis, land use and social acceptability parameters to define the extent of
possible wind farms at regional and country-wide scales.
• Grid connection variables to select sites with the best available connections to the grid, and
subsequently more preferable project economics.
• Our modelling has been successfully used by Genesis NZ to evaluate new sites for wind
farm development and determine possible turbine sites at current projects.
3. Spatial modelling with ArcGIS and Arc-SDM
Basic single layer modelling using interpolation
to estimate values between known point data.
Multi-variable models: Fuzzy logic, neural
networking, and weights of evidence modelling.
Illustrated maps that highlight important
features or values.
Type of Spatial modelling:
Tools used: Spatial Analyst Fuzzy Overlay (ArcGIS 10) and the
Spatial Data Modeller toolbar (Arc-SDM)
4. • Mineral Exploration
• Canadian Geological Survey
• New Zealand Ministry of Economic Development
• Numerous private companies in New Zealand, Australia, Asia,
Africa, the Middle East, Europe and the Pacific.
• Agricultural and Environmental
• New Zealand wide & Hawke’s Bay region grape growth potential
• Rare Powelliphanta land snail habitats (New Zealand)
• South Island Alpine Gecko habitat modelling for New Zealand
Department of Conservation
• Energy
• Geothermal modelling in the Northern Territory of Australia
• Wind Energy Potential in New Zealand,
Australia and Argentina.
Modelling Successes
5. Fuzzy Logic approach to Wind farm prospecting
• The Fuzzy Logic model is knowledge-driven: it utilises expert inputs from wind turbine
engineers and relevant spatial data to investigate unexplored areas
• We take this information and create a series of binary or multiclass predictive maps in
a GIS that represent the suitability of each theme identified
• These maps are weighted based on their relative importance and statistically
combined to create a predictive model of suitable wind farm sites
• Transparent modelling process – at any stage you can see what is going into the
model and how the inputs are affecting the final result
• Being the method data-based, we can prospect over large areas (country’s/states) to
give a comprehensive understanding of the wind farm potential
• We then focus in on smaller regional areas to do more detailed modelling that will
help to define the extent of the wind farm and turbine layout
• This process will save time and money often spent investigating unsuitable sites
6. Data Analysis Example: Wind Speed
• Creating predictive maps from
wind speed dataset created by
Aurecon Mesoscale modelling
• Reclassifying the raster based on
its wind speed values
• Assigning weights to the predictive
map classes (SDM Categorical &
Reclass tool)
Class Description Weight
1 Not Windy 0.1
2 Slightly Windy 0.2
3 Moderately Windy 0.7
4 Extremely Windy 0.8
5 Very Windy 0.9
7. • Key parameters for energy capture and turbine loads:
Wind speed distribution
Turbulence
Inflow angle
• Wind resource can be assessed with mesoscale modelling
• Turbulence and inflow angle influenced by local and surrounding terrain:
Slope near turbine:
• Influences inflow angle
• Extreme slope can cause flow
separation
• Construction difficulties
Complexity of surrounding terrain:
• Influences inflow angle
• Infulences turbulence
Wind and Terrain parameters
Terrain effects:
• Ambient turbulence from upwind
terrain features
• No higher ground in main wind
direction and no more than 1/10
height to distance ratio
Alignment:
• Slope at turbine location facing main
wind direction
• Local terrain perpendicular to main
wind direction
8. - Slope at site of turbine
- Maximum rate of change
from one cell to the next
- Average of slope over a
wider area
- Defines areas of smooth and
rough terrain
- Alignment of site with the
main wind direction
- Elevation of terrain in the
main wind direction
- Identifies sites with no
obstructive upwind
topographic features
Terrain Layers
Wind and Terrain
model scheme
9. Social Acceptability Parameters
• Other non-technical parameters related to infrastructure, social, land use and
environment need to be considered for showing currently suitable areas for wind
farm development.
• The layers analysed depend on the data availability, model scale and the area
studied. Some of the themes used in our models are:
Land use and protected areas
Distance from built-up areas
Population density
Distance from waterways
Elevation
Proximity to roads
• Layers are weighted and combined
using Fuzzy Operator AND
Social Acceptability
model scheme
10. Transmission grid modelling
• Accessing a suitable transmission grid is becoming a key factor in wind farm
development, especially in areas where the topography is not a constraint
• Availability of data can strongly influence the model:
Obtain, create and verify power line and terminal stations datasets
Find information and attributes (i.e. voltage, capacity, marginal loss factor etc…)
Create predictive layers based on the information gained
• Basic layers used are:
- Distance from lines and terminal stations
- Density of lines and terminal stations
• We can add other layers to the model depending on the information obtained for
the area.
• The layers are weighted and combined with the Fuzzy operator SUM
11. Transmission grid modelling
• The aim of this preliminary grid model
is to rank existing wind and terrain
targets in order of probability of gaining
a good connection to the grid.
13. Target Generation and Ranking
• The model results can be reclassified to produce specific target areas that are ideal for
the development of a wind farm.
• Using the modelling statistics the targets can be ranked in order of suitability for wind
farm development.
• This allows for the development of a pipeline of projects where the highest ranked
targets are worked on first because they have the best chance of success based on the
parameters used for the modelling.
For Example:
Target Name NZ Ranking Wind Class Post Probability
Kini Creek 1 1 0.9987
Manuherikia River 2 1 0.9875
Mount Haszard 3 1 0.9865
Simpson Creek 4 1 0.9712
Cave Hill 5 1 0.9701
Logans Mistake 6 1 0.9525
Collins Creek 7 1 0.9417
Poplars Fan Stream 8 1 0.9408
Camden 9 1 0.9289
Grandview Mountain 10 1 0.9114
Malrick 11 1 0.8852
14. Model Stage 1
• Model in 2 phases
• Stage 1a - Wind speed with simple country wide terrain analysis to
show areas that are technically feasible for wind farm development
• Stage 1b - Include environmental, social, land use and infrastructure
layers to show areas that are currently suitable for wind farm
development
• Produce a list of ranked targets based on modelling statistics
Model Stage 2
• Advanced prospect scale modelling can be completed over targets
identified by the New Zealand scale models
• Detailed wind speed and terrain layers ( 1km wind speed resolution, 50m terrain
resolution, ridgelines etc.)
An Example from New Zealand
15. Stage 2 Results –
Wellington
Class 1
High Probability
Low Probability
Moderate Probability
Model Output
16. Model Target Areas
- Wind speed and terrain
parameters
Stage 2 Results –
Wellington
Class 1
17. Remodel to exclude built-
up areas and unsuitable
land use (e.g. ecological
areas)
Stage 2 Results –
Wellington
Class 1
Suburbs
Conservation Reserve
18. Stage 2 Results –
Wellington
Class 1
Identification of unsuitable
areas not excluded by the
model
• Alternative land uses
• Weather radar stations
• Environmental or
archaeological concerns
• Consent issues
19. Stage 2 Results –
Wellington
Class 1
Model validated by
existing and planned
turbine locations
asic statistical gridding allows you to predict unknown values from within a single layer such as topography, geochemistry, hydrology or climate data. However the real power of GIS is when spatial modelling is applied to combine information from several layers to predict outcomes based on probability
Wind speed is the most important criteria when assessing the suitability of a site for a wind farm. We use wind speed and direction data from mesoscale modelling that is developed from three-dimensional models that simulate airflow over complex terrain and use high-quality historical meteorological observations to characterise the long-term wind distribution of a particular area. For our model wind speed data is analysed and classified into ranges suitable for the different classes of modern turbines and regional economic constraints of the study area.
Wind speed is the most important criteria when assessing the suitability of a site for a wind farm. We use wind speed and direction data from mesoscale modelling that is developed from three-dimensional models that simulate airflow over complex terrain and use high-quality historical meteorological observations to characterise the long-term wind distribution of a particular area. For our model wind speed data is analysed and classified into ranges suitable for the different classes of modern turbines and regional economic constraints of the study area.
Terrain analysis and wind speed data can be used to determine the technically feasible locations for wind turbine operation…
…elevation as higher altitudes not only create difficult construction conditions but reduced air density causes lower energy production…